mosfet gm matching under D1 IV operating conditions - diyAudio
Go Back   Home > Forums > Design & Build > Equipment & Tools

Equipment & Tools From test equipment to hand tools

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 7th April 2011, 05:04 PM   #1
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
Default mosfet gm matching under D1 IV operating conditions

Using difference cct (opamp) to multiply 1khz mVrms from mosfet VGS for accurate gm matching

Hi there guys, i've had a long road putting together a test rig to accurately measure transconductance in circuit for properly matched devices for the D1 mosfet IV stage group. i have come up against several roadbocks to get this done along the way.

i'm fairly sure i'm on the right track here, but I could use some feedback as its stil just an idea, although most is built and i have all the parts to finish it. i realize i could use a dual opamp instead of 2 singles and a much lower supply voltage for the opamps, but i wanted to have this a bit overkill and these psu voltages are handy to me. i'm open to ideas, as long as they dont mean starting again with a heap of different components, i'm sure there are lots of things i havent thought of and i look to those with real experience. the operating point for the DUT is set with 45vdc supply voltage, 21vds and iD of 120mA. the 1khz sine input signal is adjusted to 3vrms across R1 for each fet

So heres where i'm at now, tapping the 1khz AC voltage at the gate and source and feeding this to an AC coupled instrumentation amplifier with unity gain input buffer and 2nd stage 100x differential voltage gain. result given by measuring accross the output with the HP3400A at its 3v setting; dividing the result/100 before feeding it to an excel calc to get gm. directly AC coupling across gate to source is not an option unfortunately; but if it works i do expect better results this way anyway

I thought about just using the difference amp with high value feedback resistors, but this would add current noise and even then wouldnt be ideal; matching impedances would be problematic and load the D1 cct down too much, so settled on the below schematic.

this is the first time i have designed a cct from scratch and drawn anything more than a scribble on paper, its drawn in Adobe illustrator, so please excuse the non-standard symbols.

the DUT (FQA32N20C) cct and the instrumentation amp cct will have a common star ground as a copper clad board, which is also tied to shield on the bnc for the HP, but fully independent power supplies and transformers. even thought about battery power and a shielded box for the diff-amp


Click the image to open in full size.

issues/steps previous to this

1. was power supply related, i couldnt use the transformer i had for the circuit due to the regs for the test rig supply being LDO's and not coping with the extra voltage.

2. we experienced the largest and most destructive flood in living memory here in Queensland

3. after i had finished the vds matching and moved forward into gm matching, i found the amp i planned to use to pump up the 1khz sine to 3vrms used virtual ground and this messed with the readings, so i built a mono Cmoy on steroids with bipolar low noise regulated supply to rectify this

4. i found that none of the 3 meters i was planning to use produced accurate enough ac millivolt rms range down low (17.5-20mv max gate->source) to differentiate an already fairly tightly bunched batch of fets, or even to give correct readings. all other readings to set the operating point and even measuring VDS they are fine

5. i found that the HP3400A analogue voltmeter i bought to finish it, although AC coupled could not be used at anything more than fullscale for the measurements, due to the possibility of an amplified DC transient killing the input stage. only the 300vac setting is possible, utterly useless for reading millivolts. this last one was really frustrating!!! also meant that placing it across anything above ground was a nono, something i'm used to doing without thinking with my Dmm

thought about doing just that and using the full-scale (-1vdc at fullscale 300vrms) DC output at the back, then feeding the result to an opamp and using one of the other meters connected to that reading dc, found that the error% on the DC conversion of the HP was far too high to realistically be of any use for such sensitive readings

so after working through all of this, i went back to the drawing board, knowing that i had all the gear i needed when coupled with the considerable parts bin i have here to breadboard up cct's to modify, or filter readings and new found confidence from trouble shooting all of this.

thanks for reading, sorry for the long winded post, hopefully i made vyself understood, appreciate any help you can give me.
Attached Images
File Type: gif D1-t-conductance-w100x-dif-.gif (32.2 KB, 255 views)

Last edited by qusp; 7th April 2011 at 05:10 PM.
  Reply With Quote
Old 7th April 2011, 05:47 PM   #2
diyAudio Member
 
Join Date: Nov 2010
Location: N.E. PA
As drawn, one input on your differential is AC coupled and the other is DC. Any offeset, along with the high gain you have, will cause issues of dynamic range or worse. If DC is killing your meter, this is not helping. Also you could capacitor couple the output to eliminate this issue, but not deal with the input offsets.

paul
  Reply With Quote
Old 7th April 2011, 06:26 PM   #3
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
thanks for the reply Paul,

the sine input is AC from an amp with vanishingly low offset and taken behind a cap from the other side any DC would be coming from (the DUT side) and i matched the same value of the cap (10uf) at the input of the other side, i had meant to pose this question though, so thanks for bringing it up. the 10uf cap at the sine input to the D1 proper (directly after where the output is tapped from, not the one on measurement apparatus tacked onto it) is there to protect the voltage source from DC leakage, so i hoped its also able to protect the input to the instrumentation amp

the inputs of the HP 3400A meter are also Ac coupled already so no need to add caps at the output (though the schematic is too blurry to tell what the value is), those caps in circuit are there for good measure (or not so good in combination?) no worries to add another cap there, but i want to be sure that this is in fact balancing it up, rather than unbalancing it.

the only DC that i'm worried about is leakage from any possible transient being too large, i had thought of also putting 100k to ground from each to shunt leakage to ground, but I worry about the impact on the D1 cct and DUT

I would prefer to put some small caps in the feedback loop to form an RC HP filter and get rid of DC, but i'm not quite sure what value that would have to be exactly, I only need 1khz and above, as i'm not interested in the total Rms heating value including the bias

Last edited by qusp; 7th April 2011 at 06:38 PM.
  Reply With Quote
Old 7th April 2011, 07:03 PM   #4
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
oh i see, u were thinking of transients from the VS side? i planned to turn this on first, but the supply voltage is much lower on this side as well, the transients and leakage i'm worried about are from the D1 cct, its DC leakage and its much higher supply voltage that is above the max for the opamps in the dif amp.

regardless something to keep in mind and possibly try out both ways is simple enough, sounds like the basic idea is sound though. i'm wiring this up deadbug and protoboard above solid ground plane, so adding a cap is no dramas

Last edited by qusp; 7th April 2011 at 07:06 PM.
  Reply With Quote
Old 9th April 2011, 04:28 AM   #5
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
also the operating point is 24vdc across r1, not 21 as on the pic, 21vds though
  Reply With Quote
Old 20th April 2011, 05:57 AM   #6
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
well i havent attached to the matching cct yet, but tested in isolation fed with the same ac signal on both inputs at a couple different voltages to check for linearity.

fed with 3Vrms on both inputs to check for error, i got the same in as out of the input buffers, measured on the inverting and noninverting pins of the 2nd stage diff amp and at the output measured with the hp3400a, i got ~3mVrms with 100x gain applied, so an error of 100000:1 or 0.001% so pretty happy with that. i ended up having to traffo couple the ipad as dc was sneaking through on the ground connection, even though its dc coupled and using its output ground as reference it has zero offset, when connected to the cct it caused ~-80mv offset when referenced to power supply ground and this caused errors in the measurement through what i guess was a ground loop of sorts.

signal source was from 'audio tool' from performance audio running on my ipad and fed via the lineout with a 600:600 line traffo from budP of onetics and amplified with what is basically an AC powered mono cmoy with the dual rails decoupled to the hilt.
Attached Images
File Type: jpg diff amp overkill above.jpg (322.4 KB, 210 views)
File Type: jpg diff amp overkill close 2.jpg (261.3 KB, 202 views)
File Type: jpg diff amp overkill underside close 2.jpg (284.0 KB, 198 views)
File Type: jpg totally perforated mosfet rig.jpg (312.3 KB, 44 views)
File Type: jpg traffo LOD.jpg (280.9 KB, 45 views)
File Type: jpg totally perforated.jpg (328.5 KB, 44 views)
File Type: jpg diff amp overkill and sine amp focused.jpg (330.8 KB, 39 views)
File Type: jpg diff amp overkill very close.jpg (311.1 KB, 53 views)
  Reply With Quote
Old 20th April 2011, 06:09 AM   #7
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
for those interested, input opamps are lme49710 running on 10v rails and each rail decoupled with 100uf pana fc, 4.7uf x7r ceramic and 100nf polypropylene. second stage is an lt1028 also running on 10v rails decoupled with 180uf sanyo oscon sp, 100nf rifa smd pps caps and 220pf cde silver mica. the 'ground plane' is made with several layers of copper shim/foil and the rails are 8awg goertz copper ribbon. gain network is comprised of plain vanilla 1% metal film resistors matched by hand and using 2 in parallel for each position, i ended up with better than 0.01% matches. overkill for sure, but i didnt want to take any chances this time and its all from my parts bin anyway
  Reply With Quote
Old 20th April 2011, 04:16 PM   #8
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
btw i realize the decoupling/reservoir stack on the lt1028 would probably have been better up the other way around, with the faster caps at the bottom with shorter lead length to the rail and ground, but it was just easier and more stable (physically) this way.

Last edited by qusp; 20th April 2011 at 04:19 PM.
  Reply With Quote
Old 5th May 2011, 02:41 PM   #9
CeeVee is offline CeeVee  Portugal
diyAudio Member
 
CeeVee's Avatar
 
Join Date: Dec 2006
Hi qusp,

I'm at the stage of build that i just need to put in the transistors and see if it runs.

your circuit is just what we need....

but i have a couple of questions for you:

What specific film caps did you use for coupling the two stages ?

Were you able to use the HP3400A ( i also have one ...) after putting them in ?.

Thanks for your commitment.
  Reply With Quote
Old 5th May 2011, 07:46 PM   #10
diyAudio Member
 
Bksabath's Avatar
 
Join Date: Dec 2007
Briliant idea
going to give it a go and report back time permitting.

I am not familiar with OPA is 0R0 on the gain brances ?

LT1028 I know a bit beter if DC offset is a problem trim pot to adjust works vell you could reduce resistor noise by going down to 33r and 3k3
LT likes to run hot and you have very large impedence on load (The meter)
But this is just to get the least noise so no mather realy.

What about themperature as VGS change a fair bit if you have a few to test sink and work shop temperature will go up in time.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Aleph P and D1 Analog IV stage ctong Pass Labs 9 14th January 2011 03:33 PM
An IV amp using the BS170R mosfet gaetan8888 Pass Labs 2 13th November 2010 02:58 PM
D1 dac, D1v3, Pass D1 with jFET IV tritosine Swap Meet 0 19th May 2010 03:41 PM
mosfet i/v conversion from D1 with pcm1704 schiller Digital Line Level 4 15th April 2009 09:26 AM
any chance to remoove coupling cap from Pass D1 IV stage elviukai Pass Labs 6 25th July 2007 06:21 PM


New To Site? Need Help?

All times are GMT. The time now is 09:41 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2